Privacy and visitor analytics
This site records sender IPs, broad browser families and referral sources for basic public-page analytics. It does not use marketing cookies or device fingerprinting for this feature.
Aggregate statistics include the public page path, page language and load time; observatory visits are reported separately from the home page. Page paths exclude query strings and fragments. Valid campaign names and referring domains may be recorded as traffic sources.
Raw IPs and public page-load sequences are restricted to the site owner for up to seven days. One-way HMAC identifiers of IPs are kept up to 30 days to estimate distinct IPs.
Do Not Track and Global Privacy Control are respected. Forms, messages, phone numbers, payment details and educational answers are not recorded in this analytics report.
Shared IPs, VPNs and dynamic addresses mean distinct IPs do not equal distinct people.
Chatting with Tira
Tira is the site’s AI guide. When you send a message, its text and recent conversation, page language and path without query parameters, public site information and a summary of your feature access are sent to OpenAI to generate a reply. Phone numbers, card details, wallet balances, login codes and payment receipts are excluded from this automatic context; please do not type them into chat either.
Chat is kept in the current browser session to continue across pages and is removed when you clear it to start a new conversation or change accounts. KeyhanGate does not persist chat transcripts on its server and requests responses with response storage disabled. OpenAI processes messages under its own data policies. Loading a page or seeing Tira’s greeting does not itself send a chat message to OpenAI.
Narrator, Tira and experiment voices
These voices are AI-generated. The server may send public tutorial, constellation, object and fixed sample scripts to OpenAI in advance and retain the prepared audio for all visitors. If an explanation is not ready yet, choosing Play or tapping a talking object sends its script to produce speech. Starting a tutorial with narration enables speech for its chapters. Background preparation does not automatically play audio in your browser.
If you choose to hear a Tira reply, that reply’s text is sent to produce speech. Personal reply audio is not stored in the shared server cache, and playback stops when you sign out or switch accounts. Narration does not request microphone access or record your voice. The provider processes text under its own data policies.